The company has 10 years of distribution experience. It is a high-tech enterprise integrating system integration, IT consulting and providing professional IT services. It is the core agent and business partner of Huawei, Lenovo, Dell, and IBM. We look forward to cooperating with you. We can customize freely according to your needs!
| Model | Custom/OEM |
| Form Factor | 4U rack server |
| Processor | Support 2 Huawei Kunpeng 920 Max models |
| Memory | Supports DDR4 |
| Local Storage |
Supports 8/12 3.5 storage disk slots (compatible with 2.5) Supports 8 U.2 NVMe Supports 2 M.2 slots (2280/22110, optional SATA/NVMe) |
| Networking |
Support native extension of Kunpeng module with 4 100G, 8 1G, or 16 10/25G networks. Support other domestic PCIe/OCP network standard cards. |
| PCLs | 15 standard PCIe slots (2 built-in) 1 OCP3.0 slot |
| GPU |
Supports 8/10 dual width full-size GPU cards Supports 12 single width full-size GPU cards |
| Manage |
Onboard Huawei Hi1711BMC module, supporting IPMI, KVM, virtual media and other functions. Rear I/O supports one RJ45 1Gbps dedicated management network port. |
| Safety | Optional support for TCM |
| Port |
Front end: VGA * 1, USB 3.0 * 2 Rear: VGA * 1, COM * 1, USB 3.0 * 2 |
| Size | Width 448mm x Height 175mm x Depth 822mm |
| Operation temperature |
Operating temperature: 5 ºC - 35 ºC Storage temperature: -40 ºC - 65 ºC |
| Humidity |
Working humidity: 8% to 90% (without condensation) Storage humidity: 5% to 95% (non condensing) |
| Operating system support | Euler, Kirin/Other |
Pure Chinese made rack mounted 4U-8GPU artificial intelligence AI server, supporting two Kunpeng 920-Max/920 Smart modules, supporting 8/10 dual width full-size GPU cards or 12 single width full-size GPU cards, providing over 1TB of video memory capacity, supporting various GPU cards based on Ascend 910/310 chips, as well as other domestic GPU cards. This product is suitable for training large language models, fine-tuning models, high concurrency inference, and building large-scale knowledge bases HPC, Application scenarios such as simulation rendering.